Oddbean

The tweet is from csuwildcat (known nostr hater) and it says "Today we released the foundation of our vision for decentralized, serverless apps, DWeb Nodes. DWNs are personal datastores that store/sync your data across your devices & remote instances. To expand availability, Block, Google, & DIF have launched a free remote DWN, open to all:" Then it displays this image: https://i.ibb.co/ynwLS8Z/dwn-image.jpg

Then csuwildcat continues in a followup tweet: Unlike public feed protocols, you can build any type of app using DIDs & DWeb Nodes that cover the full range of use cases, from social media, to permissioned document collaboration, to encrypted personal medical data. Get started using the Web5 JS lib: And he links to these docs: https://developer.tbd.website/docs/web5/build/decentralized-web-nodes/web5-connect/ Finally, he says: Next up, I will be launching a DWeb app I've been writing that exercises the full range of DWeb capabilities. It's a consumer app I hope devs and average users alike can try to provide us feedback on all aspects of the Web5 platform. Join us in building DWeb apps of your own!

Having a hard time following the docs for a simple understanding, and I’m curious how this works: What is your first impression of this and how easy do you think it could be for the average person to run and work a DWN without a web server or port forwarding etc?

I haven't read the docs yet so my only impressions are these: - csuwildcat is knowledgeable/credible on the topic of decentralization - he says dwn relays are better than nostr relays for most purposes - I want to check it out to see if I agree but I am busy right now

the concept (from earlier documentation) is based on DWN nodes that store the history of the identifiers. biggest problem is the size of the nodes and therefore the number of participating nodes will limited. i can not find the latest system requirements. the original concept is developed when csuwildcat was working at Microsoft in decentralized identifiers.

you are too focused on ports or web servers. A web server is just an executable, and opening ports or using free http tunneling service is the easiest part about hosting your own data. And the simplicity you win using holepunching , you lose on the other side because it doesn't work in browsers, unless you then use a relay, but then that is what an http tunneling service is. finally, people don't want to host their own data, not because they physically can't open a port, but because they then need to manage electricity, connectivity, handle any bugs, upgrades, database migrations, backups etc. sure for popular stuff like movies on Bittorrent it is magical, but if you are going to be mostly the only provider of the data, might as well just use a hosting service you trust. otherwise you are just pretending.

I am a bad person to ask because I am pro-nostr and am therefore unlikely to understand his criticisms properly. My impression of his opinion, which is probably tinted with my nostr-colored glasses, is that he has two main criticisms: - nostr's design and usage patterns discourage key rotation/recovery - nostr has no good methods for creating/storing non-public content